pandafish-dt-7b / README.md
ichigoberry's picture
Update README.md
10bf0c4 verified
metadata
tags:
  - merge
  - mergekit
  - lazymergekit
  - CultriX/MergeCeption-7B-v3
base_model:
  - CultriX/MergeCeption-7B-v3
license: apache-2.0

pandafish-dt-7b

pandafish-dt-7b is a dare_ties merge of Experiment26-7B and MergeCeption-7B-v3 using LazyMergekit by mlabonne

πŸ’¬ Try it

Playground on Huggingface Space

⚑ Quantized models

πŸ† Evals

Evals from the Nous Benchmark suite:

Model Average AGIEval GPT4All TruthfulQA Bigbench
AlphaMonarch-7B πŸ“„ 62.74 45.37 77.01 78.39 50.2
Monarch-7B πŸ“„ 62.68 45.48 77.07 78.04 50.14
🐑 pandafish-dt-7b πŸ“„ 62.65 45.24 77.19 78.41 49.76
MonarchPipe-7B-slerp πŸ“„ 58.77 46.12 74.89 66.59 47.49
NeuralHermes-2.5-Mistral-7B πŸ“„ 53.51 43.67 73.24 55.37 41.76
Mistral-7B-Instruct-v0.2 πŸ“„ 54.81 38.5 71.64 66.82 42.29
OpenHermes-2.5-Mistral-7B πŸ“„ 52.42 42.75 72.99 52.99 40.94
pandafish-7b πŸ“„ 51.99 40 74.23 53.22 40.51

🧩 Configuration

models:
  - model: yam-peleg/Experiment26-7B
    # No parameters necessary for base model
  - model: CultriX/MergeCeption-7B-v3
    parameters:
      density: 0.53
      weight: 0.4
merge_method: dare_ties
base_model: yam-peleg/Experiment26-7B
parameters:
  int8_mask: true
dtype: bfloat16

πŸ’» Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "ichigoberry/pandafish-dt-7b"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])